\documentclass{article}
\usepackage{array}
\usepackage{amsmath}
\usepackage{amsthm}
\usepackage{xspace}
\usepackage{bm}
\theoremstyle{plain}
\newtheorem{assumption}{Assumption}
\usepackage{amsfonts}
\usepackage{enumerate}
\usepackage{minitoc}

\usepackage{multicol,multirow,booktabs}
\usepackage[flushleft]{threeparttable}
%%Colorful links
\usepackage[colorlinks=true,linkcolor=black, citecolor=black, urlcolor=black, anchorcolor=black]{hyperref}
%% EPS Graphics with pdfLaTeX
\usepackage[pdftex]{graphicx}
\usepackage{graphicx}
%appendix
\usepackage[title,titletoc,toc,page,header]{appendix}
%floating
\usepackage[section]{placeins}
\usepackage{float}
%horizontally oriented pages
\usepackage{lscape}
\usepackage{nth}
\usepackage{siunitx}
\usepackage{mathtools}
\usepackage{dsfont}
\usepackage{amsfonts}
\usepackage{dsfont}
%il pacchetto geometry riduce i margini laterali
\usepackage{geometry}
%note in figures
\usepackage{caption}
\newcommand{\note}[1]{\footnote{\doublespacing #1}}


%%%NEW COMMANDS
\newcommand{\foo}{\hspace{-2.3pt}$\bullet$ \hspace{5pt}}
\makeatletter
\newcommand*{\defeq}{\mathrel{\rlap{%
                     \raisebox{0.3ex}{$\m@th\cdot$}}%
                     \raisebox{-0.3ex}{$\m@th\cdot$}}%
                     =}
\makeatother





\begin{document}

\section{Main paper}

 \begin{figure}[h!]
 		\caption{Estimates of PE in simulated data. True PE is 0.3.}
 		\label{figure:sim_result}
 		\centering
 		\includegraphics[scale=0.3]{simulation}
 		\caption*{  
 		\scriptsize Note: Vertical red lines indicate the optimal bandwidth by Calonico et al. (2014).  Linear model estimated with OLS with standard errors adjusted for heteroskedasticity.
 		} 
 	\end{figure}



	\begin{table}[h!]
		\centering
		\caption{Replication of Lee et al. (2004)}
		\scriptsize
				\input{table1_col123.tex}
				\input{table1_col456.tex}
				\input{table1_col789.tex}
				\begin{tablenotes}
				\scriptsize	 \item Note:  Linear model estimated with OLS without controlling for the margin of victory. Robust standard errors in parenthesis. Bandwidth = $2$ percentage points. 
				\end{tablenotes}
							
		\label{table:leeetal_ext}
	\end{table}



		\begin{landscape}
		\begin{table}[h!]
			\centering
				\caption{Roll-call voting.}
			\scriptsize
			\resizebox{22.5cm}{!}{
				\begin{threeparttable}
					\input{table2_upper}
					\vspace*{0.3cm}	
					\input{table2_lower}			
					\begin{tablenotes}
						 \scriptsize	\item Note: 
						 Linear model estimated with OLS controlling linearly for the margin of victory on each side of the threshold. Standard errors clustered at the electoral district. Bandwidth =  0.183 selected using the method by  Calonico et al. (2014).
					\end{tablenotes}
				\end{threeparttable}
			}
		
			\label{table:ada_anderson}
		\end{table}
		\end{landscape}


\begin{landscape}
		\vspace*{\fill}
		
		\begin{table}[h!]
			\centering
			\caption{Campaign financing.
			}
			\scriptsize
			\resizebox{22.5cm}{!}{
				\begin{threeparttable}
    	\input{table3.tex}			
					\begin{tablenotes}
						\scriptsize	\item Note:  Linear model estimated with OLS controlling linearly for the margin of victory on each side of the threshold. Standard errors clustered at the electoral district  in parenthesis. Bandwidth = 0.09 selected using the method by  Calonico et al. (2014). 
					\end{tablenotes}
				\end{threeparttable}
			}
			
			\label{table:money}
		\end{table}
			\vspace*{\fill}
		\end{landscape}

\section{Appendix}


		\setcounter{table}{0}
		\renewcommand{\thetable}{B\arabic{table}}
		\setcounter{figure}{0}
		\setcounter{equation}{0}
		\renewcommand{\thefigure}{B\arabic{figure}}	
		\renewcommand{\theequation}{A\arabic{equation}}	


 	\begin{table}[h!]
 		\centering
 		 		\caption{Summary statistics - simulated data.}
 		\scriptsize
 		\resizebox{\textwidth}{!}{
 			\begin{threeparttable}
 				\vspace*{0.5cm}	
 				\input{sim_sum.tex}
 				\vspace*{0.7cm}	
 				\input{sim_corr.tex}
 				\vspace*{0.7cm}	
 				\input{sim_corrcov.tex}	
 				\begin{tablenotes}
 					\scriptsize	\item
 				\end{tablenotes}
 			\end{threeparttable}
 		}
 		\label{table:sim_sum}
 	\end{table}	

 \begin{figure}
 		\caption{Estimates of \textit{partisan effect} in simulated data. Alternative simulation. True effect=0.3.}
 		\label{figure:sim_result_alt}
 		\centering
 		\includegraphics[scale=0.35]{sim30}
 		\caption*{\scriptsize Note: each panel reports estimates of the \textit{partisan effect} $\alpha_1$ from a different sub-sample of 50 election-years, with different ratios of democratic years reported below. Estimates and 95\% confidence interval are plotted against the bandwidth used. Vertical red lines indicate the optimal bandwidth by   Calonico et al. (2014). Estimation by OLS, and standard errors adjusted for heteroskedasticity. The ``short" model include as regressors: $D_{it}$, the margin of victory, and its interaction with an indicator for observations to the right of the threshold. The ``long+year-FE" model control for both majority status and year fixed effects. The true \textit{partisan effect} is equal to 0.3 .} 
 	\end{figure}


\begin{table}[h!]
		\centering
		\scriptsize
		\resizebox{\textwidth}{!}{
			\begin{threeparttable}
				\input{leeetal_sum.tex}
				\begin{tablenotes}
					\scriptsize	\item
				\end{tablenotes}
			\end{threeparttable}
		}
		\caption{Summary statistics of key variables in Lee et al. (2004)}
		\label{table:leeetal_sum}
	\end{table}	
	


 	\begin{table}[h!]
 		\centering
 		\caption{Replication of Lee Moretti and Butler (2004): additional controls}\label{table:leeetal_ext}
 		\scriptsize
 		\resizebox{\textwidth}{!}{
 			\begin{threeparttable}
 				\input{tableB3_upper.tex}
 				\vspace*{0.2cm}	
 				\input{tableB3_middle.tex}
 				\vspace*{0.2cm}	
 				\input{tableB3_lower.tex}
 				\begin{tablenotes}
 					 \tiny	\item Note: OLS regressions without controlling for the margin of victory. Robust standard errors in parenthesis. Observations included only if the margin of victory is between $\pm 2$ percentage points. Controls include dummies for age, gender, relative who served, secondary education, college, last occupation and military service.
 				\end{tablenotes}
 							\end{threeparttable}
 		}
 	\end{table}
	

	\begin{figure}[h!]
		\caption{Replication of  Lee Moretti and Butler (2004): bandwidth robustness}
		\label{figure:leeetal_result}
		\centering
		\includegraphics[scale=0.34]{leeetal.pdf}
		\caption*{\scriptsize Note: The three upper panel report RD estimates of the \textit{partisan effect} and 95\% confidence interval plotted against the bandwidth used. Vertical red lines indicate the optimal bandwidth by \cite{Calonico2014}. Estimation by OLS, and standard errors adjusted for heteroskedasticity. The ``short" model includes: $D_{it}$, the margin of victory, and its interaction with an indicator for observations to the right of the threshold. The ``long+year-FE" model also controls for majority status and year fixed effects. The elect component is the product of the estimates in the central and right upper panels. The \textit{affect} component is the difference between the estimate in the upper left panel and the \textit{elect} component.} 
	\end{figure}



\begin{landscape}
		\begin{figure}[h!]
			\caption{\textit{Partisan effect} and \textit{majority status effect} on conservativeness in roll-call voting.}
			\label{figure:ada_anderson}
			\centering
			\includegraphics[scale=0.45]{ada_anderson.pdf}
			\caption*{\scriptsize Note: RD estimates of the \textit{partisan effect} and 95\% confidence intervals plotted against the bandwidth used. Outcome variable: adjusted ADA score. Lower values of the ADA score represents more conservative roll-call voting; higher values, more liberal roll-call voting. Vertical red lines indicate the optimal bandwidth. Estimation by OLS, and standard errors clustered at the district level. The ``short" model includes: $D_{it}$, the margin of victory, and its interaction with $D_{it}$. The ``long+year-FE" model also controls for majority status and electoral term fixed effects.} 
		\end{figure}
	\end{landscape}
	


	\begin{table}[h!]
		\centering
		\scriptsize
		\resizebox{\textwidth}{!}{
			\begin{threeparttable}
				\input{ada_anderson_sum.tex}
				\begin{tablenotes}
					\scriptsize	\item
				\end{tablenotes}
			\end{threeparttable}
		}
		\caption{Summary statistics, U.S. House electoral terms 1947-2008}
		\label{table:ada_anderson_sum}
	\end{table}	



		\begin{landscape}
			\vspace*{\fill}
		\begin{figure}[h!]
			\caption{\textit{partisan effect} and majority status con campaign financing}
			\label{figure:nongroup}
			\centering
			\includegraphics[scale=0.45]{nongroup.pdf}
			\caption*{\scriptsize Note: Outcome variable are campaign funds from non ``investor" donors in thousands of 1990 U.S. dollars. RD estimates of the \textit{partisan effect} and 95\% confidence intervals plotted against the bandwidth used. Vertical red lines indicate the optimal bandwidth. Estimation by OLS, and standard errors clustered at the district level. The ``short" model includes: $D_{it}$, the margin of victory, and its interaction with $D_{it}$. The ``long+year-FE" model also controls for majority status and electoral term fixed effects.} 
		\end{figure}
		\vspace*{\fill}
	\end{landscape}
	

	
\begin{table}[h!]
		\centering
		\scriptsize
		\resizebox{\textwidth}{!}{
			\begin{threeparttable}
				\input{moneynongroup_sum.tex}
				\begin{tablenotes}
					\scriptsize	\item
				\end{tablenotes}
			\end{threeparttable}
		}
		\caption{Summary statistics, U.S. House electoral terms 1979-2006 from Fouirnaies and Hall (2014).}
		\label{table:nongroup_sum}
	\end{table}	
		



\end{document}
